I have an argument with my dad regarding energy consumption of a gas-fired wall radiator home central heating system. Their current setup involves no thermostat and the boiler deciding when to fire based on some kind of curve calibrated to use an outside temperature reading, in order to keep the house at a set, constant temperature 24/7. I claim that they might save energy by instead getting an indoor thermostat which will instruct the boiler to only fire when needed based on the actual indoor temperature readings, and set time-based programs which keep the temperature lower at night or when the occupants are away at work.

He claims that all the boiler installers told him (and that it's the "common sense") that overall gas consumption is higher if he lets that happen - i.e. if he uses a programmable thermostat, as letting the house fall in temperature at some times and then take it up to target temperature supposedly consumes more energy than just heating it to the target at all times. I think that's ridiculous and in fact against basic laws of thermodynamics, unless the efficiency of the boiler is vastly lower at higher power outputs (which I highly doubt it is).

I kinda feel like that "common sense" is based on the olden times when we used a manual coal-fired furnace, where the efficiency was indeed vastly lower when the temperature not kept constant (as the furnace would need to be started up from scratch and then set to full roar for a while before reaching its peak efficiency operation a few hours later). Gas is different. It starts heating on demand within seconds.

Anyway. I was wondering if there is something I'm missing there. Can you please let me resolve this argument?

  • But the boiler does cycle on and off, right? It would have to, unless it has some means to adjust, in a liner fashion, it's heat output. Otherwise, if the boiler was sized to run continuously at, say, 35 deg F, it would need be able to keep the house warm when the temperataure drops to 20 def F.
    – SteveSh
    Nov 6, 2022 at 12:09
  • Also, the US Department of Energy (DOE) used to publish guidelines on this topic. I have not looked there for many years.
    – SteveSh
    Nov 6, 2022 at 12:11
  • 2
    "Standby losses" (from airflow through the boiler when off) can be significant. But those are only likely to be less overall if the boiler/burner is "modulating" (can turn down the flame, not just on/off.) As for the argument, I suggest that if you are not paying their gas bill, you just stop arguing about their choice at their house.
    – Ecnerwal
    Nov 6, 2022 at 14:24
  • 2
    Related: Do smart or programmable thermostats actually save money?. Note that for a furnace, the answer is "yes". Here we're talking about a boiler.
    – LShaver
    Nov 7, 2022 at 12:52
  • 1
    Google should show that where the choice is keeping things slow and steady at an "optimum" rate, or accelerating half the time and decelerating for the rest, the acceleration model wins out. That applies to everything from the speed of ground vehicles or space-ships to, yes, the temperature of water in boilers… I'm sorry I can't cite instances. Nov 8, 2022 at 1:03

3 Answers 3


Your dad is likely right (as usual? :-)). With modern equipment, slow but steady wins the race.

I assume we're talking about a modern high efficiency condensing gas boiler. If that's not the case and you still have an old-fashioned conventional unit, please do us all a favour and have that replaced first thing in the morning. (If you happen to be in Europe, that replacement will likely pay for itself before this winter is over.)

As the name suggests, "condensing" boilers achieve their super high efficiency by cooling the combustion products down below the dew point, so that the vapour produced by combustion condenses to liquid water, releasing a lot of heat in the process. The exhaust from such a boiler then has roughly the temperature of human breath. In order to do this, the boiler has to work with return water temperatures no higher than about 50 degrees Celsius.

Thing is, radiator systems were commonly sized to old non-condensing systems which work with much hotter water. Because the power output of a radiator scales with water temperature, running the radiators colder means you have to keep them running for much longer to deliver enough heat to the room.

At the same time, modern boilers use burners capable of throttling all the way down to perhaps 20% of their maximum power. At install time, the boiler is programmed with a curve telling it exactly how much heating power is needed to compensate the losses of your house at given outdoor temperature, and the boiler uses it to produce exactly enough heat to keep interior temperature stable.

This is useful because the boiler is typically the most efficient in this low-power, deeply throttled regime (where it produces the least amount of combustion gases, so they stay longer in contact with the heat exchanger and have enough time to condense very thoroughly).

In contrast to this, old boilers used to know only two states: "off" and "full blast". Because "full blast" had to be powerful enough to keep you from freezing when the temperature goes below -20 °C once in a half century, the only option to stay comfortable at other times was to use a thermostat to switch the boiler on and off.

Unfortunately, bang-bang control mixes poorly with the modern kind of boilers with modulated power. Remember that the boiler is calibrated to deliver just enough heat to make up for losses? Guess what happens if the termostat is configured for a big nighttime set-back, so the heating cuts out in the evening. Your house gets cold overnight and when the termostat turns heat on in the morning, the temperature stays uncomfortably cold until noon or later as there's very little excess power to raise it. Depending on how smart your boiler is, it might realize that this is going on and turn into a high-power mode (or you will get unhappy with how cold your house is and switch it by hand to a more aggressive curve). Either way, with the boiler asked to deliver lots of heat in a short time, water temperatures will go way up, above the condensation point, significantly compromising efficiency. (This effect is even more prononunced with heat pumps: Those excel at the low-temperature slow-n-steady mode, but if they realize they can't keep up with the demand, they'll go into full panic mode and turn on fallback resistive heating elements, giving you 1) a nice warm and cozy home, and 2) a heart-stopping power bill).

By the way, there's nothing wrong with thermodynamics: Domestic heating has much more to do with human perception than thermodynamics. With radiators staying only mildly warm to touch all day, you will likely feel more comfortable even at a lower room temperature than with bang-bang control. That's because when the heating cuts out completely, you lose two things: 1) radiant heat transfer from the radiator, and 2) air circulation. With radiators off, air in the room will stratify (warm air will end up under the ceiling while the floor will be much colder, especially around windows). Humans feel much colder if their feet are cold, even if the average temperature in the room is still okay.

  • 9
    If you're in Europe, get a heat pump, get out of the gas game altogether. Gas suppliers in Europe have issues. Nov 6, 2022 at 21:19
  • 4
    @Harper agreed, but the electricity side of things ain't looking any better.... Might be better off this particular winter having both options available. Though with cop up around 4.5-5 heat pumps win the numbers' game. Cheapest heat to get is the one outside.
    – Stian
    Nov 6, 2022 at 21:43
  • 6
    Yeah at COP 5 it's twice as efficient to use PutinGas in the power plant and a heat pump at the house. Technology Connections did a whole video on this. youtube.com/…. Newest heat pumps also work colder, so they don't need emergency heat anymore. On electricity, that can improve. Gas will never improve. Nov 6, 2022 at 21:52
  • 5
    @Harper-ReinstateMonica That's all true,but right now,at least in some parts of Europe,heat pump installers all have backlogs all the way to next summer,and the availability of heat pumps themselves is "call us in January, we'll hopefully give you an estimate then". Plus getting a heat pump requires a several times bigger investment than replacing a gas boiler,so many simply don't have the funds right now.
    – TooTea
    Nov 6, 2022 at 21:57
  • 6
    But as long as the heater never leaves the efficient "condensing mode", having lower temperatures at night should still use less fuel. Just don't let the house get cooler than it can compensate for when the heat throttles up again in the early morning hours so that the house reaches an adequate temperature when you get up. The dip cannot be as deep and long as it could be with a "bang bang" heater, but there can be a dip. Nov 6, 2022 at 22:31

I would not go with slow and steady, however I would also not go with a complete shut off at night. Assuming a temperate climate (so a heating season and two transitional periods), and a modern gas fired boiler with said thermostat, it does pay off lowering the thermostat at night (to, say, 17C).

The important variable to keep in mind here is the outdoor temperature, which fluctuates day and night.

Heat flows faster when the temperature difference is higher, so at the same insulation value, more heat flows out through the walls and windows of your house at night. If you drop that indoor temperature by a few degrees at night (when the outdoor temperature is a lot lower), gas consumption over the heating season will be lower. And in the morning, in an hour or so, the house will be back up to operating temperature.

For anyone wandering in on this question: note that both heat pumps and very well insulated homes (say, energy label A) are different cases.

The US Department of Energy has written a helpful primer on programmable thermostats (including a youtube video): https://www.energy.gov/energysaver/programmable-thermostats


On conceptual model I use in thinking about similar issues is a poorly-insulated volume of air surrounding a block of metal.

The specific heat of the metal is much higher than the specific heat of the air.

If the air is the same temperature as the metal, the system will lose heat from the air to the outside. So you have to pump in enough energy to make up for the heat lost from the air (minimal, because air has a low specific heat), but don't need to warm up the metal.

If you use a scheme that lets the temperature fluctuate, the metal is going to bleed and eat heat to stabilize it. For example, if you set it up so that when the air temperature drops 5 degrees your device kicks in and warms it back up, you will have a huge, huge energy swing from also letting the metal heat and cool.

That model translates to your situation too. You probably don't have a giant block of metal in your house (or maybe you do? Rock on if so!), but the walls and internal structure of your house all have a much higher specific heat than the air inside it.

That does NOT mean that constantly running the furnace/boiler is most efficient, because the boiler has its own constraints. But it does help me think about one of the factors in the overall balance.

(Quick story — A long time ago I heard that my local university had turned off air conditioning in the dorms over spring break to save energy, and had ended up spending more energy because of it. That was surprising to me, though it's hard to argue with empirical evidence. This was the model that helped me understand it)

  • The metal loses less energy to the air when it is cooler. If you let it cool instead of keeping it warm for a certain time period, it loses less energy in that time period, and you need to put in less energy to warm it back up. It may look like you are putting in more, because you are putting it in all at once, instead of spreading it throughout the time period. Nov 8, 2022 at 14:21

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.