I am tasked with optimizing the heating system of a quite large building (a cooperatively run house where four families live together in separated living areas). It currently has two gas boilers working on four separate circuits (all joined together of course when they return to the boilers). I am doing this on a volunteer basis and at the moment so I'm not very knowledgeable on the subject.

The planned (not entirely by me, I kinda received it as something that should be realized) setup is the following: enter image description here

The main goal is to integrate several inputs and needs (in green) via a control setup (red) consisting of a RasPi with Home Assistant and an RSV63 cascade controller. The RasPi would control whether the separate circuits are sending heated water, and the RSV63 would control the boilers (on/off and heating water temp as well). My main issue with this is that it seems to me that the RasPi and the RSV63 are in interference with each other in a bad way, and it might make more sense just to use one "top-level" controlling unit. Is this right? Or am I missing something?

  • 3
    I'm not sure I agree to close the question. While it may be more of a design question, it's no more so than someone asking about how big a beam needs to be.
    – SteveSh
    Commented Nov 16, 2022 at 15:25
  • 3
    And a "large cooperatively run house" is still a home - no implication of a management company, just the folks that live there trying to make things work more efficiently. Presumably the "are in conflict" assumption could be addressed by communication between the two devices.
    – Ecnerwal
    Commented Nov 16, 2022 at 15:41
  • 2
    Why do you think the RasPi and the RSV63 are going to interfere with each other? The RSV63 is going to make sure the hot water is at some minimum temperature,and the RasPI is going to turn on the individual pumps as needed. Note the later could be done with just 4 individual thermostats, no RasPI/software needed.
    – SteveSh
    Commented Nov 16, 2022 at 15:57
  • @SteveSh make that a answer
    – Traveler
    Commented Nov 16, 2022 at 16:10
  • 1
    @Harper-ReinstateMonica you've got to heat your water somehow and any sensible design would set the valves up so you could do that independently of space heating. So any losses compared to a more conventional system are only due to differences in efficiency which could go either way. Perhaps you're misunderstanding the bypass that modern boilers need - the pump runs for a couple of minutes after the boiler stops burning gas, then is turned off by the boiler. The boiler stop burning gas when the last of the room/hot water tank thermostats stops calling for heat.
    – Chris H
    Commented Nov 17, 2022 at 6:39

6 Answers 6


Don't consort with fools

There is a spectrum in intelligence. The stupidest people buy commercial off-the-shelf solutions because they can't design a system. The midrange intelligence can design a system, so they do because it makes them feel like the smartest. Problem: becoming absolutely responsible for the system working, because no contractor will touch it with a 3041mm (10 foot) pole for fear of liability. Being wise to that problem, the smartest people buy commercial off-the-shelf solutions because they need it to actually work and because if it isn't working and they can't be bothered dealing with it, they want to Just Call Someone and it's fixed.

In that light, the "middle intelligence" solution is actually the dumbest.

Seriously. Raspberry Pi, are you kidding me?

So, don't you touch this project with a 3041mm (10 foot) pole. This person is going to fail badly, and when the rest of the greater family is freezing their tails off and turns desperately to you, you want to say

I recommend a commercial solution.

Now I notice you're concerned with heating this time of year (so not Australia or South Africa), and you don't seem American or Canadian, so that leaves folks in a bloody mess for gas supply. At least you have a high efficiency condensing boiler. Unfortunately this requires new thinking if you actually want to cash in on that efficiency.

There can be a place for homebrew automation, but it should act like a "trim tab" on an airplane - capable of fine tunes but not capable of breaking the system if due to a malfunction it went "hard over" in the worst possible direction. That's my opinion. A bad design would be one where the automation needs to hold a valve open for there to be any heat at all.

BANG-BANG is dead. Long live condensing!

The old thinking was that you had a thermostat that waited until it was 1 degree below setpoint, then BANG! It turned on the furnace, which ran until 1 degree above setpoint, then BANG! It turned off the furnace. Heat level is controlled by duty cycle - on a warm day it runs 5% of the time, during the worst case conditions it runs 90% of the time. This Bang-Bang "thinking" is a disease that blocks understanding of your system.

Because - you have a condensing unit. Your unit takes methane or propane (already a gas) and mixes it with oxygen from the air (already a gas) and makes CO2 and water vapor (already a gas). We then have an exhaust-water heat exchanger and it's so-and-so efficient. But a funny thing happens. Remember that water vapor? To boil a kilogram of water takes 630 watt-hours or 2150 BTU to overcome latent heat of vaporization. We didn't pay for that, the components were already vapor - but if our heat exchanger is good enough, we can take that energy. And that's what a condensing unit does that makes it so darned efficient. This means either an absolutely huge heat exchanger, or a normal-sized one run at much lower power/flame. The lower the flame, the more efficient the unit runs.

So to score that efficiency, "BANG-BANG" must be thrown away. The new paradigm is the system runs continuously at the lowest power setting which will heat the building, adjusting the power setting to suit the demand. That is why your unit has an outdoor temperature sensor. It can adjust its power setting to correlate to the amount of heat your house needs to "hold even" with losses through insulation, which are proportional to temperature difference.

And this is where your buddy's plan flames out. It has solenoid valves or pumps that it wants to turn on BANG or turn off BANG to suit resident need. That's what the smart sockets are for. That is totally incompatible with the design concept of this condensing boiler.

What you really need is "trim" adjustments on each unit, or each radiator, to slightly increase or decrease their flow resistance, so they get more or less flow relative to the other units. It might be viable to BANG-interrupt 1 unit at a time, so sometimes 3 are running instead of 4, but that's as far as it can be pushed.

Of course, the appeal of BANG-BANG is that when you make a temperature change, the system goes BANG! full-on, or BANG! full-off, until it reaches the newly commanded set point. Your demanding customers grew up expecting that kind of BANG-BANG "fast as I can get there" responsiveness, and aren't going to take well to the slow and easy transitions made by fine adjustments. This also "takes some getting used to" for Americans accustomed to time-of-day thermostat adjustments.

Given a group of irresponsible people who don't get it, just forget hydronic and go mini-split

Or keep the hydronic in the background for baseline "get it to 15C/60F" and then have them run their individual mini-split heat pumps to fine-tune the temperature to the degree they'd like. That would help stir the air anyway; a major factor in comfort is the heat separating with the hot air in the ceiling and the cold air at the floor.

Bonus points: the mini-splits provide air conditioning. And and that fits hand in glove: a mini-split right-sized for supplemental heating will be right-sized for air conditioning in most places.

Splitting bills on a large house makes no sense.

You have shared walls, and those walls are not a source of heat loss. Occupants with a lot of shared walls have easier heating bills than people with a huge amount of outside-wall or roof exposure. And that's completely unfair, because it has no reflection whatsoever on the square footage (meterage?) of the dwelling. The people on the top floor just get mauled, because they are losing heat through the walls and the roof. Whereas the guy with a middle unit with other occupants above, below and to the side is a Free Rider, getting most of their walls heated for free by other occupants, and only has loss out their relatively small outside wall surface.

The only fair way to split the heating bill, then, is by total square footage (floor space in their private zone) with some proportional division of commons spaces, perhaps in the same ratio.

As said, if some want the thermostat set higher than others, they can use supplemental heating of their own - cheap resistive heaters (install baseboards, don't let them use unsafe plug-in heater since a house fire affects everyone)... or as I proposed they can buy a mini-split heat pump.

Insulation comes first

Once you realize that leakage through exterior walls is your actual enemy and not each other, suddenly insulation rises to the top of the shared interests. Insulation is cheaper than fuel because you only have to buy it once, and not from "unfriendly nations". If it wasn't for your communal situation, everyone would agree insulation is Job One.

  • 6
    A huge +1 on pushing for a commercial solution. Any sort of home-brew solution will be nothing but a headache down the line unless you are signing up to do all the future maintenance, in which case it will be solely your headache.
    – Llaves
    Commented Nov 17, 2022 at 16:53
  • An oil-filled portable heater is safer than portable resistive heaters. But they are heavier, slower and more expensive. Commented Nov 17, 2022 at 17:12
  • 1
    You got it all wrong, both poles should be 7 millimetres longer. ;-)
    – TooTea
    Commented Nov 18, 2022 at 10:20
  • 1
    @TooTea Harper does a great service for this site, please do not shame him for his pole being too short. Commented Nov 18, 2022 at 11:33
  • 1
    @SteveSummit, I don´t know what they would be called in the US, but on this side of the pond they are usually referred to as "modulating thermostat", or "smart thermostat" if they have additional features like a WiFi connection. They use a protocol like OpenTherm to communicate with the boiler. Commented Nov 20, 2022 at 8:35

The biggest worry I see is that modern boilers expect to controls their pumps, so the pump runs on until there's no risk of getting close to actual boiling in the (misnamed) boiler.

A solution closer to typical conventions would be to have a single pump connected to the boiler, replace the pumps you've shown with motorised valves (2-port zone valves are designed exactly for this) and somewhere have a bypass path. In a single dwelling this is often a radiator in the hall, or a heated towel rail, that doesn't have a thermostatic valve. Its own valve needs to be no more than partially open. A small (undersized) radiator in any communal space would serve this purpose nicely, but it could just be a pipe at the manifold where the separate circuits diverge. In that case, again, it needs a slightly-open valve (though actually 2 in series provides better control). Fully open it's a sort of short circuit. If there's no communal space to heat with a bypass radiator, an automatic bypass valve will do the job for you.

With more programming you have 2 further alternatives:

  • A fifth zone valve that opens when the others are all closed (just before closing the last one).
  • leave the last valve open when the boiler is cut off.

If it wasn't for the two boiler setup, I'd do without the RSV63 and control the boiler using relays connected to the RPi (which can easily be your clock/programmable thermostat).

A word of warning (though you sound like you have a good idea what you're doing). Central heating control uses mains logic, i.e. you've got 230V (at least where I live) on your control lines. Beware of this during both design (include appropriate isolation) and construction, also note that common room thermostats are mains powered and can't be used with low voltage DC. You may need to choose carefully or isolate another set of inputs to your RPi.

  • I have given some thought to replacing my own central heating controller with an RPi, but it's a slightly different problem (I'll probably do it when I get solar because I'll have some things to control to optimise my usage and the RPi could do that too.
    – Chris H
    Commented Nov 16, 2022 at 16:31
  • 1
    One thing slowing me down is the need to have something that can be maintained by a future user - I wonder hop that would work here
    – Chris H
    Commented Nov 16, 2022 at 16:38
  • 2
    Room thermostats in the US are typiically 24V.. interest that they're wall voltage elsewhere; good reminder to not assume, especially across different building codes.
    – keshlam
    Commented Nov 16, 2022 at 17:19
  • 1
    Automatic bypass valves exist as an alternative for the solutions mentioned (e.g. like this one). Commented Nov 18, 2022 at 15:46
  • 1
    Thanks @SteveMelnikoff. I've never needed one myself but now you mention it it rings a bell. I'll add it in.
    – Chris H
    Commented Nov 18, 2022 at 16:31

Why do you think the RasPi and the RSV63 are going to interfere with each other? The RSV63 is going to make sure the hot water is at some minimum temperature,and the RasPI is going to turn on the individual pumps as needed. Note the later could be done with just 4 individual thermostats, no RasPI/software needed.

I'm not sure what the connection between the RasPI and the RSV63 buys you.

  • Note the boilers will expect flow after the demand for heat stops, to stop the heat exchanger overheating (at least mine does), so you would need to figure out how to keep at least one pump flowing rather than directly switching all of the pumps by local thermostats.
    – Chris H
    Commented Nov 16, 2022 at 16:30
  • So the system I'm coming from (might want to look for an earlier question by OP) has a continuous loop for the boiler pump, so hot water is continuously circulating in this loop.
    – SteveSh
    Commented Nov 16, 2022 at 16:41
  • yes, that's the sort of thing. There isn't one shown in the diagram and it would be a bad idea to not have one
    – Chris H
    Commented Nov 16, 2022 at 16:44
  • My old heating system in my house (2 zone HWBB) had a single pump and 2 valves. The pump would run when either valve was open - calling for heat in that part of the house. I do not think it had a 3rd path for the condition you mentioned - to stop the heat exchanger for overheating. Maybe it just delayed the closing of the valve and turning off the pump for a short time after the boiler had shut down?
    – SteveSh
    Commented Nov 16, 2022 at 16:49
  • 1
    The controller may well have kept open a valve and kept the pump running after the call for heat stopped. I think my old system, replaced about 10 years ago, was similar, because the installer had to rewire the pump into the new boiler instead of the controller. I had to help him figure out the old wiring (partly because it needed an assistant, and partly because it was a bit odd for a gas fitter to work out)!
    – Chris H
    Commented Nov 16, 2022 at 16:55

I am tasked with optimizing the heating system of a quite large building (a cooperatively run house where four families live together in separated living areas)

based on what you've posted thus far, you cannot optimize without at least these

  1. building floorplan, how many stories?
  2. map of the hydronic heating circuits
  3. what is the heat load calculation "manual j" of the building, is the current boiler sized properly and what actually is the correct boiler size?
  4. how is domestic hot water to be handled, by the same boiler?
  5. you said 4 families in same house, how is billing handled?
  6. location

you are thinking of using a raspberri pi ? There are proper heating controls & electronics from Taco and Caleffi to name a couple that solves boiler control; you trying to rig a raspberry pi will get you punched in the face in the middle of winter when the heat is out for N people under a single roof even if you were all one family.

I recommed you check out Taco website (taco university youtube) as well as Caleffi. My guess would be a single natural gas boiler having a primary/secondary loop setup with each hydronic zone having it's own circulator... kinda basic https://www.caleffi.com/sites/default/files/file/hydraulicseparation-tr07.pdf; and if you do have multiple boilers then have a hydraulic separator. And maybe go with one or more on-demand hot water (tankless) heaters for domestic hot water, especially if there are summer months where there would be no call for building heat. The optimized as in most efficient would be a modulating condensing natural gas boiler for heat, and then if domestic hot water piping allows have separate on-demand tankless heaters for each family to meet each's domestic hot water needs. Everyone taking a shower at 7am in winter you would have to size for. A multiple boiler setup I suspect one would be small to meet the minimal heat needs so it is not oversized, and then the second boiler would be the same size or larger that would operate simultaneously when heating needs are at their highest and whenever can't be handled by the one smaller boiler. A lot of this is described and talked about in educational vids and pdfs from Caleffi and Taco.

does this setup make sense for controlling heating - no, not really mostly because not enough information based on your request of "how to optimized a heating system". I suggest you read as much as you can from Taco and Caleffi, and others, to understand what's involved in hydronic heating and to understand what is currently in your building and what is possible going forward in your building (is it just 4 hydronic zones? if so a taco controller easily handles that and is what you would want... for starters to look into).


As currently phrased this is an X/Y question. You are asking "could this work" without telling us what the ultimate goal is or how the pi will be programmed, which means we can't tell you whether this is a good approach or "will work" in practical terms, only that it is possible to make it do something not-completely-unressonable.i strongly suggest taking a step back, starting with the reak goal, and asking for ideas on how to solve that.

But I'll try anyway.

If I'm reading your design correctly, AND reading your mind correctly, the pi is essentially replacing the thermostats, presumably using some networked sensors as input.

That could work, but introduced more failure points and I'm not convinced it's a net improvement.

If you're just trying to apportion usage for analysis and/or billing, it seems simpler to just have the pi monitor the pumps.


I have a commercial "smart" system, which can handle multiple zones, has variable valves on radiators and has on/off switches for wet underfloor heating. The first versions of their product were actually based on a Raspberry Pi (although they never really said as much).

The point I'm trying to make is that a smart multi-zone heating system is actually very hard. Using on/off zones as you are makes it simpler (because you know when they're actually using heat or when they're not), but you lose out on being able to heat to the right temperature without over (or under) heating - which the "smart" radiator valves can do for you (they use some clever maths and a dose of "learning" to figure out how far to open a radiator inlet to heat the room to exactly the right temperature).

The problem with these "smart" radiator valves is they can't tell you if they "on" or "off", so it's hard to know when to turn the boiler on (or off). For that, you need a temperature sensor (eg. a thermostat). See below for that.

Where you can't use a variable valve like above, the system uses regular on/off valves - such as in wet underfloor heating. In theory you could use a variable valve there, but in practice it makes little difference and there are a few practical concerns why they aren't used.

In order to heat a UFH zone to the right temperature, you need a thermostat or similar to tell you what temperature the room is at. It's a relatively simple task to turn it off if the temperature is too high, or on again if it's too low. But... in some rooms you end up dramatically over-heating because of the residual heat in the floor after you've turned the valve off, and in others you under-heat because the room loses heat faster than your hysteresis and the UFH can react, so you end up bouncing below too cold and "just right". Doing any better than this is a hard problem.

In a radiator room though, "you" don't control the temperature - the smart valve does. "You" need to ensure there's hot water in the heating circuit, and the radiator valves open and close as they need. You clearly can't leave the boiler on 24x7 just in case a radiator needs it though. Instead, you have a thermostat in the room which says "we're at the demand temperature, so I'll turn off the boiler". Sounds simple enough, but problems come in when the radiator valve and thermostat don't agree on what the room temperature is. You end up in situations where the system turns on the boiler, but the radiator valve doesn't open. Solving for such issues, in a way that the user can understand, is actually quite a challenge.

Add to that any issues of wireless communications between thermostats, valves and other devices, and you've got a very "high maintenance" system which will need your constant attention.

As I say, I have a commercially developed product - they've solved hundreds of the problems inherent with such systems, but even still, hundreds more are left. Every single one of them results in support calls, forum questions and the occasional bad online review.

Lastly, and this has been touched on by @harper above - your boiler's efficiency. A condensing boiler is happiest when the input to it is cold enough that it has some work to do. That is, if you're regularly returning nearly-hot-enough-but-not-quite water, then it has to heat it a bit, which is inefficient because the condenser doesn't work. Thus, running your boiler when you don't really need to is expensive, but doing any better than that is another hard problem. Heating control looks easy, but is actually incredibly difficult.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.