GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Audio Play Icon Headphones Icon Plus Icon Minus Icon Check Icon Print Icon Picture icon Single Arrow Icon Double Arrow Icon Hamburger Icon TV Icon Close Icon Sorted Hamburger/Search Icon

Community and Q&A

Heat stratification in a tight house

DCContrarian | Posted in General Questions on

Here’s a building science/theory question. During heating season,  houses are often much warmer at the ceiling than at the floor. You can notice this going up the stairs. In order for this to happen there has to be a temperature difference in the air, and I would think that would primarily be due to infiltration, with very cold outside air entering the house.

In a very tight house there is less infiltration, which I think would mean that overall the air within the house would be more uniform in temperature. Absent infiltration the tendency of air in a contained space will be toward uniformity in temperature as any warmer air loses heat to cooler air.

Am I right? Does anyone have real-world experience to support or contradict my theory?

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. Malcolm Taylor | | #1


    No, you are bang on. That's one of the big advantages of well air-sealed and insulated houses - but you need both to avoid stratification.

  2. User avatar
    Robert Opaluch | | #2

    I'd sort of agree with Malcolm that a well-sealed and insulated building would have far less problem with heat stratification. However some stratification would still occur. The main difference is that with heat rising (that's always true), the heat at the top of the building is more likely to escape from a leaky building, and cold air is most likely to get sucked into the building at the bottom. Hence, heat stratification is magnified by infiltration of cold air.

    The reverse is also true in summer. If you air condition the upper floor of a building, cool air is likely to fall down stairways or other open spaces to floors below. So an air conditioner located on the upper floor of a home could help cool the lower floor (cool air falls as heated air rises), but cooling air on the lower floor would not effectively cool the upper floor. In winter, the reverse: The heat generated on the first floor would rise to the upper floor more effectively than heat on the upper floor moving to the lower floor.

    Another factor: Heat moves not only from convection (air currents, described above), but also from conduction (heat moving through objects in contact). So heat can move through walls, ceilings and floors to other rooms, but those do have resistance to heat movement (measured by R-value).

    So this issue involves insulation of the building. If a building is both airtight and very well insulated, then the internal heat will stabilize throughout the building. Typically, interior partitions, ceilings and floors (not facing the exterior) have fairly low R-value. Exterior walls, ceilings/roofs facing the exterior, and lowest flooring of the building have higher R-values. If the insulation of the building's exterior shell is very high, interior heat during wintertime will move between rooms much faster than it moves out of the building to the exterior. So heat becomes better distributed inside the building if the building shell is both air-tight and well-insulated.

    Bottom line: For a stable and well-distributed indoor temperature, seal the building very well, and insulate it very well, to reduce cold air entering (more at the bottom) and allowing heat to dissipate throughout the building evenly.

    1. User avatar
      Robert Opaluch | | #3

      For a real-world example, here's one involving air-conditioning. My elderly mother was insistent on not allowing any window air conditioner in her small Cape style home in New England. She claimed it dries out her skin. (However she sat under ceiling fans, which cool people by moving air rapidly across the skin, increasing evaporation of sweat, and drying the skin!)

      On one occasion, I visited during our hottest summer weather, and found her miserable with interior temp of 88F and high interior humidity. (A medically risky situation for an elderly person.) Since she was insistent on no window air conditioner, I installed one upstairs in a bedroom, which was connected to a 10' hallway to a steep stairway to the living room on the main floor of the home. Over a few hours, the humidity of the first floor decreased significantly, and the temperature of the living room dropped. Eventually, the temperature of the kitchen and bedroom dropped, as they were connected to the living room by a small central hallway. And eventually she realized that the house was more comfortable with an air conditioner operating, so (with some nagging), allowed the air conditioner to operate in her main floor bedroom during the day when she was not in the bedroom. (It would gradually dehumidify and cool the main floor.) However, the air conditioner in the main floor bedroom failed to cool the upper floor bedrooms and hallway.

      During the winter, the upper floors had minimal heating from the central heating system (no return air duct, and ineffective uninsulated supply ducts with long duct runs, definitely not a balanced system!) Opening the door at the bottom of the stairway would allow heat to rise up the stairway, heating the upstairs modestly. You could feel the cool air breeze exiting the door at the bottom of the stairs, despite no fans operating. So heat would rise to the upper floor moderately effectively during winter, but cooler air in summer would not.

      1. User avatar
        Michael Maines | | #9

        "The main difference is that with heat rising (that's always true)"

        That's not actually true--heat goes in whatever direction there is lower energy. Warm air, however, is more buoyant than cooler air that is otherwise equivalent. In a tight house there is very little stratification.

  3. Keith Gustafson | | #4

    I don't think you need outside air to create stratification

    But in a tight house you have less volume of less hot air from the heating system, meaning I would think less chance of stratification

    A poorly designed heating system, IE heat upstairs, might still have some stratification.

  4. Akos | | #5

    I happen to measure this yesterday. I had 0.5C between floor and ceiling peak (14'). So not much.

    In the summer with AC you do see much more though. Because there is no cooling up high, you get much more stratification. I remember measuring around 3C.

    1. DCContrarian | | #6

      Thank you for providing hard numbers!

  5. DCContrarian | | #7

    So here's the context of the question: I'm considering ceiling radiant heat for a new build. First question from the architect was "won't all the heat stay at the ceiling?"

    I think in a well-sealed, well-insulated house it won't because there will be little convection. Air-to-air conduction will tend to even out the heat in the room. To the extent there is radiation from the ceiling to objects below that will tend to warm the air from below which again will lead to evening.

    1. User avatar
      Robert Opaluch | | #8

      Radiant heat is a third type of heat transfer (other two are conduction and convection). Examples are your proposed ceiling radiant heat units, the sun shining on you, and a hot wood stove. Heat jumps through space or air, from the hot object to cooler objects. Radiant heat panels would heat objects in the room; and those objects would then distribute heat to the surrounding air by conduction and convection. So the ceiling radiant panel would not just heat the air at the ceiling as the architect believes, radiant heat would heat objects below the ceiling panel.

      Radiant heating tends to be perceived as very pleasant when the environment is cool or cold (e.g., heat from the sun on a cold day, or heat from a wood stove in a cold room). Central heating that distributes hot air to a cold room are perceived as less pleasant; Objects are still cool to the touch until the air finally transfers heat to the objects over time. Air doesn't have much thermal mass (heat capacity) so it is typically hotter than the temperature desired in the room, to move enough heat to heat all the objects in the room more quickly. At first that hot air might feel good in a cold room. But when the room is warmer, the hotter than ideal air temperature is not perceived as so pleasant. So radiant heating units or radiators are more pleasant than hot air distribution systems.

      1. DCContrarian | | #14

        Even radiators -- despite their name -- transmit heat through a mixture of radiation and conduction to the surrounding air. The heat transmitted through conduction is then distributed through convection depending on what the air in the room is doing.

        One thing I'm trying to figure out is the relative share of radiation vs conduction. Ideally there would be 100% radiation. If it's mostly conduction I would expect a temperature gradient from ceiling to floor. Very simplistically, imagine a room with a heat source at the ceiling and an insulated floor exposed below to the outside. For the purposes of simplification assume the walls have infinite insulation. There will be a temperature gradient from the ceiling to the bottom of the floor. At any given point the temperature will be determined by the amount of insulation above and below. So the temperature above the floor is determined by how much insulation is below the floor and how much insulation the air between the floor and ceiling provides (which is itself a tricky question).

        However, if radiation is a significant method of heat transfer the radiation is hitting the floor itself and warming it directly. So the relative contribution of conduction and radiation matter.

        One of my concerns is that radiant panels are typically run at lower temperatures that radiators -- maybe 110F as opposed to 180F. The formula for radiation depends on the fourth power of the absolute temperatures of the two surfaces; plugging in those temperatures and a room temperature of 70F I get about 3.3 times as much radiation at 180 as at 110. But ceiling panels could easily be triple the size of a regular radiator so I think you could get the same net radiation.

        1. Brad | | #21

          As others have said in various ways, you will, have stratification whenever there is a source of heat (heat air duct, radiator), or a source of cold (infiltration, walls, windows), and some convection, or a source of heat at the ceiling. It's an interesting question, surely someone has done some testing? There are probably some academic papers, but they are hard for laymen to find. 110F does seem kind of low to get a lot of radiation. Since it's at the ceiling you won't have natural convection. You will have conduction through the air and if that will give you a temperature gradient. You could make a rough calc with the radiative heat transfer equation, and the heat transfer parameters for the floor & furniture. My guess is that it would be similar to having a heat register high on the wall.

    2. Trevor Lambert | | #10

      We have radiant cove heaters and never noticed it being hotter near the ceiling. In fact, if anything I would say the effect is greater using the high wall mounted minisplit.

    3. Akos | | #13

      You have to first figure out why the ceiling heat makes sense.

      For hydronic, the part and labour cost is about the same but you need almost 1.5 to 2 times the surface area. I guess if you have a high temp heat source (ie outdoor wood boiler) it might be better but than you start running into the temperature limits of drywall.

      Electric would be similar (floor heat vs radiant ceiling panel), maybe cove heaters VS baseboard might make sense for interior space.

      1. DCContrarian | | #15

        Ceiling appeals to me for a number of reasons. Performance doesn't depend on the placement of furniture or floor coverings. You can run it at a hotter temperature than a floor because you don't touch it.

        Radiant floors have to operate inside a narrow window, you can't really have the surface temperature above 100F or it becomes uncomfortable to walk on. But that's not much of a delta from room temperature. If the heat loss of the room is more than calculated or the heat contribution of the floor is less than calculated, you don't have a lot of room to make adjustments once the floor is installed. With a ceiling you have more leeway.

        1. Akos | | #19

          Most well insulated houses need relatively small amount of floor heat. Even with the stricter temperature limits, if you cover the entire area, the floor will never get hot enough to feel it. For comfort, you actually only want to put the floor heat in the higher traffic areas that way you get that warm toes feel when it is cold outside.

          For example, I'm easily heating a 4ACH50 balloon framed house with just blown cellulose walls and 20 year old double pane windows in zone 5 with floor heat. You need a ridiculously lossy house to not be able to heat it within the floor temp limits.

          Furniture is not a problem. You do have to watch for carpeting though. Smaller carpets are fine, larger shaggy ones can be an issue.

  6. Andy CD Zone 5 - NW Ohio | | #11

    DC, here is some real-world experience. I live in an extremely tight SIP house in CZ 5 that's taller than it is wide. I have three floors on a 28x18 footprint, and the blower door result was something under 0.5 ACH50, below the limit of the tester's equipment. Unfortunately, I have serious air stratification issues. There's an open stairwell between the floors, so air movement is quite free to occur. In hindsight, I would have designed things differently. I should have installed more insulation in the finished basement (currently have R-15 continuous interior on walls, R-10 under slab) and I would locate my HVAC differently.

    I have two 9K minisplits: a ceiling cassette on the first floor which handles all wintertime heating, and a mini-ducted in the peak of the second-floor cathedral ceiling which is used for cooling in summer.

    On these cold-ish winter days, the basement can be up to 10 degrees colder than the upper floors, and that gradient seems noticeable in our feet as we walk on the first floor. Of course, the only active heat source for the ENTIRE house is in the ceiling of the first floor, a full 16' above the basement floor. It just can't push warm air down that far. In hindsight, I would substitute the ceiling cassette for a mini-ducted in the floor (basement ceiling) to serve both the first floor AND the basement.

    Another contributing factor (the main culprit?) is the HRV, which in my case is a relatively inefficient unit dumping barely conditioned fresh air into the basement and upstairs bedrooms. It's only 30 cfm spread over three rooms, but I'm not willing to damper off the basement, so perhaps spending up on a better HRV would have been prudent.

    1. User avatar
      Michael Maines | | #12

      Andy, I would say that a 10°F temperature difference between an unheated basement and the second floor is actually quite good. In leaky homes the difference could be like 20 or 30 degrees, or more. Without a heat source in the basement, even with decent levels of insulation, it is a heat sink--it wants to be close to the ground temperature, roughly 50-55° where you are. If it's warmer than that, and I bet it is, it's from heat coming from upstairs and miscellaneous heat loss from water pipes, water heater, etc..

      Have you considered adding inexpensive electric resistance heat in the basement to take the edge off? I'm currently designing a basement fit-out in Maine with specs not unlike yours; the heat load for the 600 sq.ft. space is only 1100 watts, and that's to keep it at 70°.

      Why are you supplying air to the basement? Even efficient HRVs supply colder air than ambient indoor conditions. I usually extract air from the basement unless it includes living space.

      1. Andy CD Zone 5 - NW Ohio | | #17

        Didn't even occur to me that 10 degrees was reasonable! #tighthousesnob. The basement gets an HRV supply because it is a bedroom (has a casement egress window.) And indeed, I have a 1500W electric baseboard heater in there, which is just fine for keeping the room comfortable when/if occupied. Its just that the COP of 1 feels like a design fail, and if the bedroom door is ajar, that "dirty" heat escapes continually up the stairs to mingle with my COP 3-4 minisplits!

        1. User avatar
          Michael Maines | | #20

          Andy, if it makes you feel better about your "dirty" heat, consider that if used intermittently, it's probably costing you on the order of $100 per year (maybe $20, maybe $200) to operate--the upcharge to go to a nice mini-split would cut that by 2/3 but would cost a few grand--hard to justify the return on investment for part-time use. If the upstairs is at the same temperature as the basement, or higher, then the bedroom heat isn't really escaping unless it's being driven by air leaks.

  7. Doug McEvers | | #16

    Andy makes a really important point. Even in a very airtight and well insulated home the distribution of heat is critical. In my house with gas forced air, the furnace runs 3 times per hour distributing heat evenly to both levels. My current temperature difference between the floor and ceiling on the main level is less than 1F. This house was built in 1978 by others but has been upgraded thermally, especially in the attic with air sealing and copious blown insulation. The temperature in the lower level is always cooler because the basement slab has no insulation, sub slab insulation in a cold climate is a must.

    I have been monitoring my neighbors house this winter for gas usage after a ceiling insulation and air sealing retrofit. Some ceilings (60%) were cathedral with the balance conventional flat ceilings over the bedrooms. The cathedral ceilings were a compromised R-22 and the conventional ceiling was about R-30, both leaked enough heat and air to cause ice dams each winter. Both ceiling areas were brought up to R-50 and made as airtight as possible. The cathedral ceilings have a dedicated airspace below the roof decking, the flat ceilings have eave and ridge venting and have been air sealed with blown cellulose added to R-50.

    Gas usage so far for the neighbor's house is 24% less this winter than before the retrofit using normalized weather data. We did a blower door test prior to the retrofit and will do another post retrofit to see what improvement was made to the ACH50/CFM50. I will be interested in the final blower door test to put some numbers on energy saved due to infiltration reduction and add this to the calculation for the thermal improvement for the ceilings. I may find the reduced gas usage is more than the the 2 improvements made would indicate using standard calculations but we shall see. Trying to understand the value of making an airtight (as possible) ceiling plane in an existing home.

    I am encouraged by this project as ice dams have been eliminated and snow now lays on the roof where before it melted due to heat loss. I firmly believe energy efficiency for a building must include an airtight and highly insulated ceiling area. The stack effect is held in check and interior temperatures will be more even throughout.

    1. Zephyr7 | | #18

      It is nice after completing an air seal / insulation project to go out on a cold morning and see the improvement in the frost pattern on the roof :-)

      I’ve seen similar results to you when I spray foamed a VERY leaky cathedral sealing in my own home. Other projects have been smaller, so less of an impact on energy use, but still noticeable. A recent renovation on my home office has made it MUCH more comfortable just because of air sealing work! It’s really worth the effort!


Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |