Hacker News new | past | comments | ask | show | jobs | submit login
Efficient Hot-Water Piping (2013) [pdf] (garykleinassociates.com)
119 points by rudedogg on Mar 7, 2018 | hide | past | web | favorite | 56 comments

So, the article correctly says that efficient copper layouts will beat pex, but in residential, those layouts pretty much don't exist in practice.

Plumbers are happy to just use whatever fittings, etc, to get pipe where it needs to go, and rarely, if ever, think about the losses caused as a result (i'm really not slagging on plumbers or the quality of work, i'm just saying the average plumber trying to get 4 homes done is trying to get 4 homes done, not sitting around calculating the most efficient layout for copper pipe when he discovers a stud in the way nobody planned for).

In the past two brand new houses i've lived in, done to the latest plumbing code, the amount of hilariously horribly copper layout is amazing.

(I'm sure anybody who does this for real has ridiculous horror stories).

Meanwhile, most pex layouts i've seen perform significantly better in practice just because they use less fittings. So the attempt to be "as horrible as copper" fails.

Also pex is a better insulator, so in the real world where pretty much no in-home copper is insulated, pex holds heat better.

So, yeah, i agree with the sentiment that well thought out, well structured copper works really well, that doesn't happen to be residential reality, and hasn't been for a long time. (it's possibly a commercial reality though)

Pex doesn't insulate as well as PPr (thinner wall) and the fittings always introduce flow resistance and produce turbulent flow since they go into the pipe. Thread fittings for Pex area usually leaky. I would never use them except for serviceable manifolds with easy access, otherwise use press fit fittings. On the other hand PPr fittings go outside the pipe, just like those for copper and they're heat fused. PPr pipes are quite freeze-prone, the walls are thick (6 mm for a DN25 pipe) and glass fibre inserion PPr pipes don't have such a huge thermal expansion as pipes w/o insertion (only suitable for up to 60..68°C = hot water). Most plastic piping is prone to thermal expansion. PPr does not look pretty so it should be probably ran through walls and floors.

Copper water pipes should always be insulated with at least 6 mm EPE sleeves or closed cell elastomeric foam sleeves (better). Even cold water pipes should be insulated at least in heated or high humidity spaces because of condensation that forms on the pipe.

"Pex doesn't insulate as well as PPr (thinner wall) and the fittings always introduce flow resistance and produce turbulent flow since they go into the pipe."

Errr, i have used systems of pex with outside fittings forever. Yes, inside fittings suck. All of the manufacturers have been sued for leaking, etc.

"Copper water pipes should always be insulated with at least 6 mm EPE sleeves or closed cell elastomeric foam sleeves (better). Even cold water pipes should be insulated at least in heated or high humidity spaces because of condensation that forms on the pipe."

Yes, they should. But they very rarely in my experience (or they did a super shitty job) are unless the house was just built and you had a good code inspector

Well, the article explicitly states one should use "outie" style fittings for PEX. Check the fittings figure on the last page; letter (C).

Exactly: Modern PEX fitting go on the outside. Where I live (Denmark), PEX is used almost exclusively for new residential installations, but I haven not seen an in-pipe PEX fitting since my parents build their house in the 80's.

This is my favorite thing I've seen on Hacker News in weeks: immediately made me interested in something I had never thought much about before, gave me the information directly and usefully and got me started on a bunch of related questions.

Speaking of which: I had always wondered why more plumbing isn't done with flexible hoses instead of this pipes. Hoses seemed like they'd be way easier to work with. Now my guess is that hoses generally have unnecessary curves no matter what and that this increases the pressure lost. Is that right?

Most materials will not last long in water, especially hot water with the various chlorinated organic compounds that result from sterilizing impure water with chlorine. Copper and CPVC are really rather special! Gaviotas had to replace the pipes in all of their aluminum-piping rooftop hot-water heaters with expensive copper when it turned out aluminum could handle chlorine but not the municipal water at their customer sites. All over the US, polybutylene piping turned out to only last about ten or twenty years due to unforeseen issues with chloramine corrosion of the PB pipes and Delrin pipe fittings, but only where they're under stress. And that's why you can't buy polybutylene pipe anywhere in the US since 1995.

And it turns out that an unexpected rupture of a water pipe can be very, very expensive indeed. Especially in a wood-framed structure with sheetrock. Especially if nobody's home.

So builders are generally very conservative with water pipe materials. Nobody wants to install the next PB disaster. Building supply companies are even more so; the PB lawsuit cost Shell a billion dollars.

Garden hoses are usually made from one of two materials: some kind of rubber (buna?) or PVC plasticized with phthalates. With hot water, neither of these will last even the 10 years that a lot of PB piping lasted, and the phthalates will leach out into the water and give it that garden-hose taste.

I don't think the curviness of hoses is generally enough to cause a significant pressure loss.

As noted in another comment, PEX is sort of flexible and hoselike, and is seeing a substantial amount of use for water piping nowadays.

Interesting fact: lead pipe which was used very early in the days of plumbing (that's where the term comes from: plumbum = lead in Latin) was installed "curvaciously", due to its relative flexibility. It also turns out that when used with hard water, it would develop a calcinous protective layer on the inside over time (and thus the amount of dissolved lead in the water would be less than naive estimates based on completely "fresh" lead pipe), making it relatively corrosion-resistant.

Aside: due to the toxicity much of the lead pipe installations no longer exist, so modern pictures of lead pipe installations are exceedingly rare, but for the curious: https://www.nachi.org/forum/f22/lead-drain-supply-2105/

The snakelike nature of lead pipe is also immortalised in early cartoons and drawings.

You may enjoy reading some early plumbing books: https://archive.org/details/standardpractic00davigoog

"As noted in another comment, PEX is sort of flexible and hoselike, and is seeing a substantial amount of use for water piping nowadays."

Yes, PEX is the flexible tubing material you are thinking should exist.

PEX and sharkbite fittings are like magic - you can plug together arbitrarily complex water circuits and layouts as simply as building with legos.

I personally use copper piping for very long-lived installations and for simple, long runs. However, anything complex at the end of the line (irrigation, pump and well work, filter assemblies, etc.) transitions to PEX and shark bites.

I keep SharkBite ball valves on hand in every pipe diameter in my house. Twice now I've had some sort of minor plumbing "emergency" (leaking appliances) that would have required shutting the water off for a day or more until I could get it repaired. Having the SharkBites on hand meant I could just stick one in-line on that branch and leave the rest of the water on while I sorted out the issue.

I had never heard of SharkBite fittings. As a minimally-competent home handyman, I thank you for putting me on to them.

Here's a link to an informative and sensible video on SharkBite... https://www.youtube.com/watch?v=1E8X1VawLeE

I have seen shark bites fail.

It is particularly problematic in inaccessible areas (hidden behind drywall).

PEX compression fittings with coupled 'sticks' of straight PEX makes runs easier (not having to straighten out the 'coil').

Also, check out the Milwaukee PEX tool. Expensive, but awesome.


In the first episode of Stranger Things, you can see the hot and cold PEX runs in the kitchen of the diner. Very progressive.

Some people are trying to make thermocoupling pipe material, so that would remove a bit of the heat part of that problem. Now would that make "reactive" fluid less problematic on the long run ?

You mean, hook up Seebeck generators to your pipes in order to harvest energy from the temperature differences? It's an interesting idea. I don't think it will help much with hot-water corrosion, and it might hurt.

First, you presumably don't want to put your Seebeck generators on the pipes that run hot water to your taps, because that would make the water coming out of the taps colder. That heat isn't waste energy; it's the effect you were trying to achieve by heating the water! You'll have to put the Seebeck generators on your drainpipes only. But usually it's the supply pipes that corrode, not the drains.

That aside, I am guessing it's going to be hard to passively pull out so much heat from the water as to make a noticeable difference in the corrosion rate. Remember that the temperature difference has to remain large enough to keep driving the current through the load. And, in order to do it, you need to hook up the Seebeck generators to the pipes — you don't want to make the pipe walls themselves out of thermocouples, because then instead of a Seebeck generator you'd have a short-circuited battery, with the water in the pipe as the electrolyte. So you end up joining four or five dissimilar materials together along the wall of the pipe — a recipe, generally speaking, for extra vulnerability to corrosion, although you can avoid that by some ways of doing the joints.

In summary, I think Seebeck energy harvesting from plumbing is very promising, but not as a way of extending plumbing lifetimes.

Generally, these days, flexible PEX (cross-linked polyethylene) piping is used for new work in most places. I don't think pressure loss would be much of a problem since the curves in a properly installed system should have a reasonable radius. The reason copper is still used seem to have more to do with inertia and a (sometimes warranted) lack of confidence in new building products.

Well, and this mess: http://www.plumbingfittingsettlement.com/FAQ.aspx

They've perhaps fixed that by now, but the reputation is there.

If you like this kind of stuff, check out 99% Invisible[0]. It's a (very popular) podcast:

> 99% Invisible is about all the thought that goes into the things we don’t think about — the unnoticed architecture and design that shape our world.

[0]: https://99percentinvisible.org/episodes/

I think the flexible hoses are very common in newer houses. The labor costs are much lower. It also performs well in cold climates. Apparently the technology started in Germany in the 80s so it is well proven. Matt Risinger, a high end home contractor has some great videos on this technology.

https://www.youtube.com/watch?v=KZX0aTvbEzo https://www.youtube.com/watch?v=OOeBJ8mDr8Q

> I had always wondered why more plumbing isn't done with flexible hoses instead of this pipes.

My understanding is that flexible hoses were not considered durable. Line your walls with flexible hoses carrying water and your insurance company is going to flip out.

Or put another way, flexible hoses didn't meet building codes.

Times change though and PEX (a kind of flexible tubing used for plumbing) is now allowed in many (all?) states.

Reminds me of this: http://www.chilipepperapp.com/

I was thinking of installing a couple of these a few years ago, but didn't and then forgot about them. It's a pump that gets installed near a faucet; you turn it on when you want hot water; it pulls water out of the hot pipe, sending it back into the cold pipe (where it circulates back to the water heater), until it senses that hot water has arrived, when it shuts off.

As I say, I haven't tried it, but maybe I will — seems like an elegant solution, and works with existing plumbing.

It sounds like a good idea although some water companies are not that happy with an end user putting water back into the supply. Technically the pipe under your sink is open all the way back to the reservoir or whatever so there’s potential for system contamination there. Systems are usually designed so that syphoning cannot happen if there’s a temporary supply problem also. This is why cisterns are fed noisily from the top rather than quietly from the bottom.

There's a one way valve on the water pipe coming into in your home, precisely to avoid contamination. It's right before the water meter.

A cistern could be fed from the bottom as long as there is a length of pipe going up (to the highest level of water) then back down. Or using a valve/backflow preventer.

A length of pipe like that wouldn't prevent the supply sucking back water though. So that doesn't solve the whole contamination issue.

Yes the pipe would need to rise from the cistern up to higher than the level of the resevoir, which would not always be practical. The resevoir would not then be able to fill the cistern over that loop :)

Right you are; I hadn't considered that if the pipe is always full of water then gravity doesn't matter (as much)

A better use for grey water is to store in a reservoir for flushing your toilets. You obviously need separate plumbing for collecting the grey water and for the toilets.

There are also recirculation pipes for lung runs of hot water piping, like hot water columns found in hotels. The return circuit doesn't need to be thick, two sizes downwards will do (for instance a DN15 pipe for a DN25 column). Both the column and the return need to be properly insulated. DHW circulation pumps are usually made of brass and come equipped with a no return valve. This layout does exactly what you describe but instead of going into the cold water pipe, it goes back into the boiler.

These kind of things are fun to think about and tricky to solve. My thermostat in my hallway for example, if I set to 69, will turn off but it still feels cold. I used an IR thermometer to see that it is 69 right at the height of the thermostat, but is a gradient down to the floor where it is like 65. So there are thermal layers in the house. I started to put a small fan on the floor blowing up at the thermostat that will both mix the layers and blow the cold floor temp up at the thermostat. It will kick on and after a bit the whole house will be uniformly toasty :) I started to think why this kind of thing was not built in, but it would require air inlets and outlets at ground level and ceiling level and two thermometers, so it starts to sound too complicated and expensive. I see why people like in-floor heating now.

This is partly solved with a right-sized furnace. A well insulated house with a huge furnace will run the furnace for two minutes an hour. Whereas a small furnace might run forty-five minutes an hour. The overall gas usage will be the same, or maybe even lower, and there is significantly more even heating & mixing.

Ceiling fans have a switch that reverses the direction it spins. Specifically for that reason, one direction is for summer and the other is for winter.

This is interesting: I never knew that was the reason. Downwards for summer to fan you, I presume..? but upwards in winter why? Or am I backwards here?

Wouldn't setting the thermostat higher to compensate for that do the same thing?

No, then my head is hot and my feet are cold :)

Tangentially related and complementary (though in a certain sense, opposite to this article), I am a big fan of the Evolve TSV shower valves.

Turn the hot water on full while you're getting ready and the shower will run until the water turns warm and then will slow the water to a small trickle, minimizing the waste of hot water down the drain with no one in the shower.

When you're ready, step in and pull the lever to restore (instantly hot) water flow.


I solved this problem for a small house w/ oversized hot water pipes by installing an endpoint electric tankless heater. It had a sensor for incoming temperature so it would apply electric heat until the hot water from the tank caught up. The initial phase wasn't as hot as the direct feed, but it made it comfortably warm (vs the very cold winter water.). Installing that was much cheaper than adding a return circulation loop - but I wish I had thought of the option dropping the pipe diameter now. Though it might have still been cheaper to install the point heater.

We're on rain water so I always cringe when running perfectly good water down the drain while waiting for it to heat up.

It would be great if there was some endpoint heater like you describe, that ran from e.g. ultracapacitors that could be trickle charged from solar.

And yeah, having just built our house, I wish I had read this article earlier and learned about using minimally sized piping! Obvious in hindsight.

For your situation, I would think a mini electric hot water tank at the endpoint could work as long as you have excess electric, you can dump it there as heat. It's much cheaper storing as heat instead of in battery or ultracap. You'd get losses from leakage, but if it's well insulated it could be worth at least checking out the math.

Ive seen some claims that solar water heating/circulation systems are now more expensive over their lifetime than collecting the solar as electric and heating water, but I've never gone through and sanity checked that calc.

Sounds good but then when that small tank ran out of hot water, I guess there would be a period of cold coming through until the main tank backed it up?

Unless the small tank was big enough for most conceivable uses - in which case, e.g. in the bathroom, it would become a large tank.

A continuous flow unit at least can do what you said in your earlier post, e.g. instantaneously heat up the cold until it starts coming through hot. That's why I was thinking the supercaps, they might provide just enough juice for those 45 seconds or whatever, and also hopefully be reliable enough for decades of use, like you would expect from a hot water appliance.

I don't understand his chart on Page 77 showing the amount of water wasted for the different layouts. He has the Zoned at 12 gallons, and the Manifold at 23 gallons. The note below says that the "estimate is based on 20 cold starts per day". I'm guessing this means total uses, with some fixtures used more than once (eg, kitchen sink 3 times and lavatories once)? And "cold start" means that it's assumed that the time between uses is such that one always needs to wait for hot water to come all the way from the tank?

The Manifold system is a "home run" with 1/2" to all fixtures except the tub, and the Zoned system is a mix of 3/4" mains and and 1/2" branches. Since the direct run should never be longer than the branched run (right?) how can the system composed of all small pipe waste more water than the system with a mix of small and large pipe? In a cold start measurement, I'd think the Manifold would have to come out ahead for every fixture except the tub, which would be a tie. And thus I'd think that any combination of fixtures would have to come out better, rather than worse.

(I'm not saying this makes the Manifold approach better overall, just that something seems off about this numbers)

You're assuming that the water cools instantly and you're always wasting the run from the boiler to the fixtures. This isn't the case: the big advantage of the zoned approach is that the trunk will remain warm if you're using multiple fixtures branched off the same trunk.

So if you have a shower while your washer is in use, you only waste that little branch of water, rather than the whole trunk (or the home run in the manifold approach). Or if you wash your hands and have a shower in the same bathroom within half an hour (for example). With the manifold approach, you lose the efficiency savings when you use the sink / washer / shower within half an hour of each other.

You're assuming that the water cools instantly and you're always wasting the run from the boiler to the fixtures.

No, I'm just assuming that the definition of a "cold start" is that the water is allowed to cool before the next measurement is taken. The alternative (which I agree is common) is a "hot start". The author defines these terms on the box on the bottom left of Page 76 (which I don't seem to be able to copy and paste). Since the chart explicitly says "cold starts", my question/complaint is that as labelled, I can't see how the chart can be correct. I agree that measuring a blend of hot and cold starts is probably a better metric to judge the systems by.

Do "smart" tap systems exist? Here's what I'm imagining: * instead of a tap being a valve that opens or closes, relieving or building pressure, a tap is a switch that signals.. * to the "smart manifold" that it should start pumping water (hot or cold) to that output. * if the tap is shut off, the water is sucked back out of that piping and recycled.

So then you would have instant hot or cold water, instead of any mixing at all.

Here are some reasons why this is a dumb idea: * can't handle branching, * pipe networks have "welling points" where water will simply remain by gravity, and no pumping will get it out.

Has anyone tried something similar?

Pumps are not good at stopping and starting. They like to start, run for at least a few minutes, and then stop. Each stop/start cycle adds wear to the pump, as the components inside it are at peak strain and friction when they speed up and slow down. Forcing a pump cycle each time someone opens a faucet would cause a lot of maintenance.

This is why wells have pressure tanks, to allow the well pump to run longer cycles, building up pressure to be released over time in the tank.

There's a second issue in that pipes don't like to be empty: exposing hot and humid surfaces to free air is a bad combination.

The stop/start issue is true, but this is also an easily solvable problem. Variable speed dc pumps are a complete commodity at this point. You see them on geothermal systems, etc.

It's just the recirculation folks are lazy AF.

That depends on the kind of pump.

Centrifugal pumps are perfectly fine with start/stop cycles and back pressure, though not with air.

Here is an "under sink" re-circulation device, instead of the "whole house" approach. Basically, put a thermocouple, solenoid-driven valve, small pump, and a coupling between the hot and cold water supplies in an under-sink cabinet at (usually) the most remote lavatory sink in the house.

When hot water is wanted, the pump pushes cooled water from the hot side into the cold side, and eventually out into the hot water heater's cold water intake. The same flow at the pump pulls hot water from the heater to the sink, and the pump shuts off and closes the connecting valve when hot water arrives at the connection.

No water lost, no re-circ except when wanted. Use the sink to set-up to bring hot water near to the shower. Locate or repeat as needed if you have multiple "cold" bathroom or kitchen locations. The reverse flow thru the cold water piping is, uhm, counter-intuitive, but appears to work.


$250, ouch. What's the payback time on that?

You can find <$100 tankless heaters on Amazon.com. I think I'd rather install one of those. Besides up front cost, you're not pushing the partially-heated water back into the cold supply, which is certainly not insulated.

People don't want air shooting out of their faucet for 5-10 seconds before water starts to come out, even if the water is the perfect temperature then. For this system to work every faucet would need an air valve out of the way that could vent the pipes before water started coming through.

Check out recirculating systems. It’s a similar idea. The modern ones have timers to run approximately when you’ll be using the hot water (ex: before morning shower).

Please don't get a recirculating system, even a smart recirculating system, if you care about the energy efficiency of your home.

It’s a trade off of power for water. If your house is entirely local solar what’s the issue?

> "If your house is entirely local solar what’s the issue?"

That is a big "if"...

Please do, but use an actual smart one. It should turn on when you start using hot water and turn off once it warms up the loop. These exist but are bizarrely hard to find.

The point of recirculating hot water is to preheat the pipes so the water is always immediately hot out of the tap, so turning it on just when you start using hot water sort of defeats the point.

Not really. You’ll still get the hot water there considerably faster.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact