Welcome! Here are the website rules, as well as some tips for using this forum.

# Really basic question about bottlenecks and flow

Member Posts: 6
This will greatly help inform the way I plan my hydronic repiping, but it is an embarrassingly stupid question:

Lets say I have a 20 foot section of 3/4 inch pipe. On one end i pump the water into it. In the middle, about 10 feet in, is a one inch length of 3/8 pipe. Now I walk over to the other end of the pipe. Is the flow limited to the flow capacity of a 3/8 inch pipe. In other words, is the flow the same as if the entire 20 foot length was 3/8?

My intuition tells me that the 3/8 bottleneck will limit the flow, but I also know assuming anything is what gets you into trouble. Thank you!

• Member Posts: 327
You are correct. The friction losses in the example pipe you mention are the sum of several items:
1. the friction in the walls of the 3/4" pipe.
2. the friction of the restriction where it reduces from 3/4" to 3/8"
3. the friction in the length of 3/8" pipe (not much in your 1" long example, but would play a role if the 3/8" section were longer)
and,
4. some very minor losses where the 3/8" expands back to 3/4".

Simple answer a restriction like will reduce the flow rate through the whole pipe below what it would have been if it was a whole 10 foot length of 3/4" pipe with no reduction in the middle.
• Member Posts: 5,795
edited December 2018
It’s like a chain that’s only as strong as its weakest link. However, it would not be the same as 20’ of 3/8” pipe. 20 feet of 3/8” would have 20 times the resistance of one foot.
Bob Boan

You can choose to do what you want, but you cannot choose the consequences.
• Member Posts: 6
edited December 2018
Ok. Currently my house has 3/8 convectors which are connected through the basement with 3/4 inch pipe. It is not parallel. So whoever did this 50+ years ago didn't understand basic plumbing. Good to know!
• Member Posts: 5,795
Is it a monoflo (one pipe) system?
Bob Boan

You can choose to do what you want, but you cannot choose the consequences.
• Member Posts: 5,995
Most of the plumbers 50+ years ago knew a lot. It is usually the guys that have touched it since that muck it up.
Post some pictures, you may have an unusual but functional older system. It is often better to understand and "roll" with older systems rather than try to modify.
"If you can't explain it simply, you don't understand it well enough"
Albert Einstein
• Member Posts: 6
edited December 2018
It is a two pipe system. But it is piped in a series. The first convector outputs to the next one's input. All of the piping up to the last 6 inches where it goes up to the 1st floor is 3/4, then it switches to 3/8. Then back to 3/4, etc, etc.

If I redo this in pex, it would be a heck of a lot easier to do it in 3/8 if there is no flow loss. In fact, it would be better since there would be less sitting water losing heat waiting to return to the boiler.

Was considering piping this in parallel with a reverse return, but now I am leaning towards a home run with a manifold, since the 3/8 is so easy to work with, and I could balance the pressure from one single location. Thoughts?
• Member Posts: 13,397
The question isn't really quite that simple. Be nice if it were. What you -- or anyone doing this sort of thing -- needs to look at is the relationship of flow and pressure loss.

Let's start with a simple situation. Suppose we have a pump which is what is called positive displacement: that is, on every revolution (or stroke) of the pump it displaces a fixed amount of fluid (water). We hook this pump up to a pipe assembly -- can be any pipe assembly. Suppose now that we make one, slow revolution of the pump. A certain amount of water will be forced out of the pump and the water will trickle through the pipe and out the other end. If we measure the pressure at the pump, it will be very low. Now suppose that we spin the pump nice and fast. Now we are moving a lot of water through the same pipe, and there will be a lot of resistance to that flow (the flow resistance of a pipe is proportional to slightly less than the square of the velocity). The pressure at the outlet of the pipe is still zero -- but the pressure at the inlet of the pipe has gone way up.

Bottom line -- positive displacement pump, you can put pretty much any flow you want to through the pipe, but you have to accept that the pressure drop through the pipe goes up more or less as the square of the flow.

Unhappily, most of the pumps we use in almost any application (note that hydraulics for machinery are a major exception) are not positive displacement, but are some variation on centrifugal pumps. They don't work the same way. For a fixed speed centrifugal pump, the flow through the pump decreases at some rate greater than linear as the pressure change across the pump increases. This varies with the design of the pump, and is shown by the pump's characteristic curve (which is usually available from the manufacturer). There is a maximum flow from the pump -- the flow which it will produce with no pressure change -- and a maximum pressure (called shutoff head) at which pressure the pump will no longer pump any water at all.

If we hook such a pump up to our pipe assembly and turn it on, we will find that it will produce one specific flow rate, which is the flow rate at which the head loss through our pipe exactly equals the head produced by the pump at that flow. If you were to draw a graph of the head loss through the pipe vs. flow and superimposed it on the pump characteristic curve, the two lines would intersect at some point -- and that point will give you the flow rate through the pipe and the pressure at the pump.

Now to go to your 3/8 inch restriction. Yes, the flow rate will change -- but so will the pressure delivered by the pump. The pump will create slightly more pressure, and the flow will drop slightly.

To extend this to your question regarding redoing the whole thing in 3/8 inch PEX vs. the 3/4 inch copper. Will the flow rate change? Yes -- if you were to use the same pump. In fact, it would probably change a lot, since your velocity would increase by a factor of 4 and the head would be almost 16 times as great. But suppose you want the same flow? No problem. Use a different pump, capable of pushing the same flow at the higher required pressure. The pressure at the pump will be a lot higher, but you'll still have the same flow.

Suppose to go further that I redo the piping as a home run system with pipe half the diameter of the original series system, and for the sake of simplicity I have four loops. Now what happens? The flow from the pump is divided among four loops, so each loop gets one quarter of the original flow. The velocity in each loop is now down to the original velocity -- and therefore the head loss in each loop is down to the original head loss, to a broad approximation. Now we look at our original pump and what do we see? It's looking at the same head loss it had before and the same flow (again, to a broad approximation) and it will happily pump just as it did before. It doesn't see anything different (short sighted of it, I agree, but pumps are pretty stupid that way).

I'm afraid that this is drastically oversimplified in a number of ways, but I hope it gets the general point across.
Br. Jamie, osb

Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England.

Hoffman Equipped System (all original except boiler), Weil-Mclain 580, 2.75 gph Carlin, Vapourstat 0.5 -- 6.0 ounces per square inch
• Member Posts: 6
Wow, thank you, thats a lot to chew on. One question though. As it stands, because all the convectors are in a series, and they are bottlenecked by a minimum diameter of 3/8, would the logic you described still apply?

If done in separate loops, they now would all have their own dedicated 3/8 lines adding up to 4x (say in this example) of the existing flow, unless I am missing something, which I most likely am.
• Member Posts: 13,397
With regard to the first point, a short section of 3/8 inch would have almost no effect on the flow. There just isn't enough of it.

With regard to the second point, I would have to make some assumptions -- but if the loop lengths were at least similar, then you wouldn't have four times the flow if the pump stayed the same -- you would have almost the same flow, just split four ways. The reduced flow in each pipe would just about compensate for the increased head loss in the smaller pipe.
Br. Jamie, osb

Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England.

Hoffman Equipped System (all original except boiler), Weil-Mclain 580, 2.75 gph Carlin, Vapourstat 0.5 -- 6.0 ounces per square inch
• Member Posts: 6
Based on what you have said, do you think its a good idea or not a good idea to go the home run route?
• Member Posts: 13,397
I'd go with the home run route. There is a lot to be said for home runs -- not the least of which is that each radiator or group of radiators or whatever can be balanced individually to suit the particular needs of the space in which it is located (balancing valves are very handy gadgets!). Were it myself, I'd probably use half inch PEX -- be sure it's the oxygen barrier sort! -- rather than 3/8 inch, since the head loss (and thus the demand on the pump) will be significantly less and it's not that much harder to work with -- but clearances may make that harder.

If there are several radiators in one space, I might get a little fancier and pipe them as a group, parallel reverse return -- but then I would surely use half inch. But that sort of consideration would depend so much on the job and accessibility and so on...
Br. Jamie, osb

Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England.

Hoffman Equipped System (all original except boiler), Weil-Mclain 580, 2.75 gph Carlin, Vapourstat 0.5 -- 6.0 ounces per square inch
• Member Posts: 13,824
Think of that small reduced section of a balance valve🤓 At typical flow rates in 3/4 tube it would have very little effect
Bob "hot rod" Rohr
trainer for Caleffi NA
The magic is in hydronics, and hydronics is in me
• Member Posts: 7,306
Back to basics. How does you house heat now? Post some pictures of the piping. To me it smells of a monoflow system. Most monoflow systems used 1/2" copper for branches but I have seen them as small as 3/8" od copper for sections that only needed a small flow.

It just depends on the btu requirements of the baseboard or convectors.

My own house originally had 1/2" copper monoflow branches wit 1" copper fin tube
• Member Posts: 6
Thank you Jamie. You have been so helpful, I have until the spring to understand all these concepts and a fresh copy of "pumping away" that needs to be read. Hopefully between your comments and the book I will avoid catastrophe.

Eberatt - definitely not a monoflow - the input and output to the convectors are not connected underneath. It goes up one side and down the other.
• Member Posts: 2,561
The other thing to check with a series-connected system is the output of the various emitters themselves. As the water temperature will be decreasing through the loop, we'll expect the output of each one to increase, to offset the drop in water temp. Something to keep in mind, if the last emitter requires e.g. ½ the flow of the first one for the same performance.