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

Why Pipes Freeze

The science behind frozen water pipes and recommended preventative measures

Copper water lines damaged in the 2021 Texas power crisis. Photo courtesy T.J. Schell, Redline Roofing and Remodeling.

As a building scientist, I never say, “I told you so.” But sometimes it’s hard.

About 20 years ago, I worked on a custom home intended to showcase natural and non-toxic building materials. One of the home’s highlights was a bathroom with stunning views of the surrounding hills. Its large, jetted tub was cantilevered over an open porch. The overhang was insulated with cotton batts and finished with tongue-in-groove pine boards.

I warned the homeowners that without a good air barrier, cold outdoor air would move through the joints in the paneling, work its way through the batts, and freeze the tub’s plumbing. My concerns were ignored. After all, I was just a local contractor, and their big-city architect assured them there was nothing to worry about. You can guess how this story ends.

Pipes freeze when exposed to sustained temperatures below 32°F. Water expands as it turns to ice; the burst pipes that result are responsible for hundreds of millions of dollars per year in insurance claims.

One of the most detailed studies I found on the mechanisms of pipe freezing and bursting is a 1996 report prepared by Jeffrey Gordon for the Insurance Institute for Property Loss Reduction. Gordon conducted experimental studies, first in a specially instrumented commercial freezer and then in a field laboratory built to simulate a residential attic. Gordon reports that pipes freeze in four distinct stages:

  1. When exposed to very cold air, the water in the pipe cools rapidly. Interestingly, the water may “supercool” or remain liquid several degrees below 32°F. This supercooled state can persist for hours.

GBA Prime

This article is only available to GBA Prime Members

Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.

Start Free Trial


  1. [email protected] | | #1


    Bill Rose has a great narrative regarding Gordon's key finding that "The pressurization of liquid water, rather than the radial expansion of ice in the frozen portion of the pipe, causes most bursting."

    Bill, like many of us, has observed that hot water lines are somewhat more likely to freeze than cold water lines. I've heard a range of theories about the reasons for this from various plumbers I've worked with, most starting with the fact that hot water is less dense than cold (which doesn't really make any sense as a cause of pipe freezing, given that the water temperature needs to get down to freezing point or below no matter where it starts). Bill's hypothesis, backed up by observation, is that toilets are, of course, always on cold water runs -- and can act as pressure relief valves, sparing the cold water pipes from bursting.


    1. Jon_Harrod | | #5

      Thanks, Paul. That is an interesting hypothesis and makes a lot of sense. Another idea I ran across is that water in the cold pipes has higher concentrations of dissolved oxygen and other gases, which can depress the freezing point.

    2. charlie_sullivan | | #8

      I wonder if if would help if toilet fill valves were explicitly designed with a pressure relief function at a controlled pressure level.

  2. Expert Member


    Fascinating stuff. The conventional wisdom to avoid frozen pipes in unheated buildings has typically been to drain the lines, but I notice that wasn't included in your list of remedies. Is that because as you (and it seems from Paul's post Bill Rose) are suggesting that it is unnecessary as long as they are not pressurized?

    1. maine_tyler | | #3

      Malcolm, what's the difference between draining the lines and "shutting off the main water supply while opening taps to provide pressure relief." Won't this drain the line, or does draining the line require somehow introducing air?

      1. Expert Member
        MALCOLM TAYLOR | | #4

        Maine Tyler,

        Using air, or sometimes all the lines are sloped with draining in mind. Shutting off the main supply can create a vacuum that keeps water in the proximate lines even when taps are open.

        Up until Jon's article I had always assumed the pipes burst where they froze as the ice expanded in circumference. That would mean you needed to drain every part of the system. But if the failures are mainly due to increases in pressure, then shutting off the supply and opening taps would suffice.

        I wish the article had come out last month. I've just re-routed all the waterlines in our community hall from where they were lying on the ground under the pier foundation up into the conditioned space. Something that may not have been necessary - although unexpected freezes would make knowing when to empty the lines (and close the hall) difficult.

        The hall was built by loggers in the late 30s, and one interesting find while crawling around under it was that the center beam, which spans over 60 feet, is continuous. God knows how they got it there from the mill in Jordan River 20 kms away.

    2. Jon_Harrod | | #6

      Hi Malcolm, I was writing mainly with continuously occupied buildings in mind, but draining down the lines can definitely make sense if the house is going to be unoccupied for a while, especially if there's an drain at a low point in the system. A common practice for three-season lake houses and forest camps in my area is to drain the lines in the fall, then fill the supply lines and drain traps with RV antifreeze. After researching this article, I'm not sure the antifreeze is really necessary in the supply lines, especially if taps are left open.

      1. Expert Member
        MALCOLM TAYLOR | | #7

        Thanks Jon. It really shouldn't be an issue here where we only get maybe a dozen days below freezing each year, but for some reason people always seem to build right up to the edge of whatever limits their climate imposes, and each winter houses, resorts, and restaurants around us are forced to close when their pipes freeze or burst. Humans are a strange species.

  3. charlie_sullivan | | #9

    I once had my heating system fail while I was away for more than a week, during a winter cold spell, and before many of the envelop upgrades I have since made to my New Hampshire house. The house was 28 degrees in the middle of conditioned space when get got home and several of the pipes (mostly copper) had burst. Even some old traps burst. Miraculously, where a pipe near the incoming city water supply burst, ice formed in a way that stopped off the supply, and water damage was minimal.

    I suspect we were very lucky and that such a occurrence is very rare, but I would be curious if anyone else has seen that happen.

  4. maine_tyler | | #10

    "Even some old traps burst"

    Yeah, so how accurate is the notion that all that's needed is pressure relief?

    1. Expert Member
      MALCOLM TAYLOR | | #11


      I think it may have been my erroneous conclusion, rather than Jon or Paul's, that pressure relief is enough. Looking back a their (and Bill Rose's) comments, they seem to say either the expansion of the frozen pipe or pressure caused by it can create ruptures.

    2. charlie_sullivan | | #12

      The traps that burst were really old and corroded, and might have failed soon even without a freeze. I don't think my experience with that proves anything.

  5. [email protected] | | #13

    As luck would have it, Bill Rose just launched a website for his (invaluable) building science musings, and here's a timely essay on freezing pipes:

    1. Expert Member
      MALCOLM TAYLOR | | #14

      Thanks Paul,

      A video showing it happen:

  6. StuSid | | #15

    Crazy thought, but if water pressure is what causes pipes to burst, could a pressure tank help absorb some of that pressure as air is much more easily compressible than water?

    if the pressure tank did provide some pressure relief during a freeze event that would only reduce the likelihood of a burst between the freeze and the pressure tank?

    I guess I am just trying to apply the same principals for why a hot water heater needs an expansion/pressure tank to allow for expansion when water is heated in a pressurized system. Maybe I am miss applying this.

    1. Jon_Harrod | | #16

      Gordon looked into this concept, building "air chambers" that essentially functioned as mini pressure tanks and inserting them into piping between the freeze zone and the fixture. He did find that these could help prevent pipe bursts under certain circumstances. As far as I know, this idea has never made it into mainstream construction.

      1. StuSid | | #17

        Thanks Jon. Super informative.

      2. Expert Member
        PETER G ENGLE PE | | #18

        Anti-water hammer devices could easily be used for this purpose, and they should work fine. The increased volume of water when it freezes is actually very small, just greater than the flexibility of the pipe can handle. This is the main reason that plastic pipes don't burst when they freeze - it's not that they are stronger (they aren't), but that they are just a little bit more flexible so that they can expand a bit without bursting.

        The problem with using any sort of expansion device is that it has to be located between the first place to freeze solid and the end of the now-closed-off pipe. If we know where the place that freezes is likely to occur, we can usually just insulate the pipe or the cold air leak, and prevent freezing in the first place. A whole-house expansion tank won't let you get away without draining pipes in a power loss situation because you would still have places that the pipes freeze first in the middle of a run, causing the dead-end sections of pipe that would still burst.

        I've done several inspections of houses that experienced freeze events during power outages, when unoccupied, or when otherwise not heated in freezing weather. There are usually many places where the pipes rupture due to many individual "first freezing" locations. In most cases, you don't even know where the ruptures are until you turn on the water again and then it's a fast game of whack-a-mole. Worse yet, copper pipes can sometimes yield plastically, forming a "blister" that is very weak, but not actually ruptured. These blisters can burst unexpectedly long after the initial event, often for no identifiable reason.

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |