GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Forecast-based thermostat?

Trevor_Lambert | Posted in General Questions on

My house has in floor heating and radiant cove heaters, is super insulated and has high solar gain. So traditional temperature setbacks and the typical “smart” thermostat or simple time programmed thermostat aren’t of much use. I’ve found that I have to change my thermostat settings based on what it’s going to be like many hours in the future. For example, if the interior winter temperature is say 18C at 7:30am and it’s going to be a sunny day, I won’t need to turn the heat on at all. On the other hand, if it’s going to be a cloudy day, then I needed to have the heat on the night before, because now it’s too late (assuming someone is home, if not then the heat can run during the day and catch up for when we get home).

Is a thermostat that connects to the internet and uses a weather forecast to set the temperature a thing? I tried searching for one, but I don’t know what it would be called. I tried the search terms “forecast thermostat”, and just got a bunch of financial forecasts for thermostat manufacturers.

I did find this one:
It mentions connecting to a forecast, but the details are sparse; seems like it’s just for extreme weather events rather than ongoing control.

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. NormanWB | | #1

    I would think a smart thermostat paired with a IFTTT script similar to this one might work.

  2. Jon_R | | #2

    A good question but also consider that if you set the slab temp to say 19-24C (depending on outdoor reset) it will automatically stop delivering heat as soon as the room temp exceeds that. So perhaps less wasted heat than you might think.

    Passive solar gain creates a rapid change in btu input - this is fundamentally incompatible with slow, high mass radiators and a desire for a steady indoor temperature.

    1. Trevor_Lambert | | #5


      I can see what you're saying, but the trade off is allowing for a wider range of acceptable temperature. I'm perfectly fine with 19, or even 18 as a low point overnight. But 24 is going to be pretty uncomfortable. I'll do some logging (I have a 5 channel thermohygrometer logger on order, just waiting for it to arrive) and see what the max temperature rise is on a sunny day. If it's on the order of 3degC, then I can probably set a timed heating schedule to be within 18-20 overnight, and 20-22 during the day, with just a minimum of manual adjustment. That would be tolerable.

  3. Trevor_Lambert | | #3


    Thanks for the reply. That's on the right track, except it's not as simple as the outside temperature. In fact, the outside temperature is almost irrelevant, it's whether it's sunny, overcast or mixed.

    This gives me an idea, however. While the forecast for the next day would be ideal, it certainly would still be an improvement to have it respond to real time solar irradiance. To that end, connecting a light sensor to some kind of smart thermostat seems like a possibility worth exploring.

    1. NormanWB | | #4

      I understand the irrelevance of the outside temp, but thought you could maybe get forecast data from someone like Weather Underground that could serve as a trigger.

      1. Trevor_Lambert | | #6

        The function seemed to be pretty specifically for temperature. I searched a bit on there and found this one:

        Trigger MESH if current weather condition changes to clear

        I don't know what the heck MESH is, but maybe this could be used.

  4. Yupster | | #7

    Sounds like you want a thermostat with a function like "True Radiant" function of the Nest smart thermostat.

    If you are not fond of the Nest thermostat, try searching for "predictive radiant floor heating thermostat".

  5. kjmass1 | | #8

    I wouldn't recommend a Nest thermostat for this application- you'll want the details that ecobee offers (can download raw spreedsheet data in 5min increments), along with multiple room sensors. Nest is far too limited for what you want to accomplish (I returned mine and made the switch).

    IFTTT is what you'll want to use connected to an ecobee. "If tomorrow's high rises above X, then set thermostat to X (setpoint, on, off)". Then set another one for the other boundary. This is also just the weather underground tie in, I'm sure there are tons of other options as well.

    There is also a UV index, current outside temp, wind, humidity, etc.

  6. GBA Editor
    Martin Holladay | | #9

    I hope that the comments you've received point you toward a solution.

    For other GBA readers, I'd like to mention that Trevor's dilemma reinforces two themes I've discussed often on these pages: high-mass radiant floors are hard to control, and traditional passive solar design principles are obsolete.

    For more information on these issues, see these three articles:

    "All About Radiant Floors"

    "Reassessing Passive Solar Design Principles"

    "All About Thermal Mass"

  7. Trevor_Lambert | | #10


    Yes, I've got some ideas. The first idea I'm going to try is doing nothing at all, as I think I am trying to solve a problem that may not even exist. Over the last couple of days I made some observations. Setting the thermostats to 19C overnight, yesterday was a very sunny day and the interior temperature only got up to 22.5C. This is pretty reasonable, and as the days are getting shorter there probably isn't going to be a day with much higher solar gain than that.

    I should point out that our floor is not designed to be high mass, it is just the concrete slab as required for structural reasons. We also don't have a crazy amount of south glazing like in the second article. It's certainly more than an average house, and we did use PHI principles in the design, the primary considerations were ambient light and views. The areas that were more affected by passive house design were limiting glazing on the other three sides of the house.

    The only things I truly regret from an energy design standpoint was having in floor heating instead of a multi-split, and not having motorized exterior shades on the east and west windows. Ok, and installing two-wire thermostat cables was a bad oversight.

    1. Expert Member
      MALCOLM TAYLOR | | #12

      "installing two-wire thermostat cables was a bad oversight."


      Was that because you don't have a neutral to run digital controls?

      1. Trevor_Lambert | | #13

        I just installed what I needed for the system I was installing (heat only radiant), and didn't anticipate future updates.

  8. paulnederland | | #11

    I am installing a new radiant floor heating system that will work together with a air-based heating system. The rad floor will have a capacity of about 1/3 the total capacity. I think I'm going to have an issue with controlling the supply water temperature, given the low heating capacity and the enormous thermal mass of the tiled floor of about 20,000 lbs.
    It will only work if I can use outside temperature forecasts and change supply water temperature settings at least 1 day before.
    Think about turning the rad floor completely off 2 days before the forecast says that outside temperature will be 70 degrees - independent of how cold it is today. Likewise, if today is 70 degrees but tomorrow it will be 55. I need to turn the rad floor ON a full day before the heat demand occurs.
    So, I need a device that pings a weather forecast service and adjusts the water supply temperature. I am sure IFTTT would work but the lack of solid security is simply unacceptable. Some hacker would be able to see that my temperature settings for tomorrow are at "away" levels; not a very comfortable feeling.
    Does anyone know of a smart thermostat that would control temperature settings for supply water based on weather forecasts?

    1. Trevor_Lambert | | #14

      I think your presumed lag times are high. 20,000 lbs isn't all that enormous. My concrete floor is about 140,000 lbs. Even with that thermal mass, there's not much heating or cooling effect beyond one day. The slab can produce noticeable effects on thermal comfort within a few hours of it turning on.

      The heating demand shouldn't vastly increase at an outside temperature of 55degF, if your house is well insulated. My heating system wouldn't even come on at that temperature.

    2. Jon_R | | #15

      A mitigating factor is that large radiators self regulate. For example, say that you predict that tomorrow will be super cold and you heat the slab to 75F. But tomorrow turns out to be 70F. Yes, the radiator will overheat the house - but not more than 75F (and usually significantly less). It also works the other way. Today is 70F so you heat the slab to 70F. Tomorrow is unexpectedly super cold - but the slab may prevent the house from dropping below 65F.

  9. paulnederland | | #16

    Trevor -
    thanks for the reply. The lag time is calculated to be about 12 hrs. Depending on heat loss during the warm-up cycle. lag time maybe closer to 16 hrs. My floor is 2.5 inch thick wood, cement board and porcelain tile. The radiant floor piping is underneath all of that so, all that mass has to be heated before the surface temperature reached the desired level. Best practice tells me that your concrete thickness above your pipes is just a few inches and so, just a part of the 140,000 pounds of concrete will have to be heated to create the desired surface temperature. Eventually, a large part of 140,000 lbs will be heated eventually. So, the same problem occurs; the heating system should probably turn OFF two or three days before the outside temperature is high enough for at least a few days. The turn ON and turn OFF decisions for the season changes should be based on at least a week's forecast.
    Of course, you could simply run the system in a continuous mode and be slightly inaccurate with the ON- OFF timing during the season changes. I've done some simulations and it would increase my heat loss by at least 5%. The frustrating part is that the data is available via internet - the device that would switch ON or OFF the rad floor based on that data is simply not available.

    Jon -
    you are right; the large thermal mass does average out some of the temperature fluctuations. However this averaging produces some loss of efficiency.
    It is like steering a 500 million pound oil tanker. The captain has to plan course changes days ahead of time to prevent fuel inefficiencies. It is counter-intuitive; the most difficult system to control is one that has a large time constant. A good map and excellent weather forecasts are essential to make the best out of burning 500 gallons of fuel an hour.
    The highest efficiency from a rad floor is only achievable if you use predictive scheduling.

    In the mean time, I found commercial controllers that perform this predictive control but at a cost of several thousands of dollars.
    I have also found a way to program a Raspberry PI computer to collect the forecast data and it should be relatively easy to turn a relay ON or OFF. Material cost is very low so, I just might start a project to built my own rad floor predictive controller!

    1. Jon_R | | #17

      Self regulation has little to do with thermal mass - it's all about a small delta-T between the radiator and the interior air temp. There is no heat flow from a 75F radiator to a 75F room.

      I agree with you that high thermal mass radiators create some (but as I explained, limited) control problems and predictive outdoor reset would help.

  10. T_Barker | | #18

    x2 Ecobee with IFTTT

  11. paulnederland | | #19

    T. - I agree that IFTTT enables a solution. It is the lack of security that keeps me from implementing it.
    There has been no reports that Ecobee has been a security concern but, still, I don't feel comfortable letting strangers know that I am not at home, or worse, give hackers the opportunity to hack into my network.
    The frustration comes from the fact that the weather data is available in a secure way and the logic is really very simple. Maybe here is a home automation hub available that allows this simple programming?

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |